A visual laser data feature compression and extraction method and system

Abstract

The application discloses a kind of visual laser data feature compression and extraction method and system, it is related to feature compression and extraction technical field, including the following steps, establish the feature compression framework based on convolutional neural network, extract intermediate feature;According to the statistical characteristics of feature, determine the non-information feature of redundancy or little information amount, and delete the feature most similar to non-information feature;Discrete cosine transform is introduced to further compress statistical redundancy, eliminate the redundancy between feature channels by feature rearrangement, feature transformation and feature quantization;In receiver, according to the feature of non-information feature subset, the removed feature is recovered.The feature in the similarity subset of the application will be deleted in the encoder, and recovered according to the non-information subset in the decoder, which significantly reduces the pressure of data transmission and storage, while ensuring the integrity and effectiveness of feature information.

Description

Technical Field

[0001] This invention relates to the field of feature compression and extraction technology, and in particular to a method and system for compressing and extracting visual laser data features. Background Technology

[0002] In the application of digital twin technology for power transmission and transformation equipment, intelligent devices equipped with vision / laser sensors have been widely used to provide various image, video, or laser data. This data is transmitted and stored to remote servers to perform various analysis and monitoring tasks, such as image classification and defect detection. How to transmit and store massive amounts of data cost-effectively and efficiently is a significant challenge. Traditional multimodal data compression methods aim to maintain good human-perceptible quality. However, with the development of cloud services and data sharing platforms, real-time status monitoring of power equipment also places higher demands on the transmission and transformation equipment's multimodal data transmission and storage.

[0003] Deep neural networks can learn feature representations by training on large-scale data without requiring hand-crafted feature extractors. Therefore, a paradigm for encoding and transmitting intermediate features can be adopted. The neural network is divided into two parts, deployed at the edge and in the cloud, respectively. The front end of the network extracts intermediate features at the edge, while the rest of the network checks task-specific accuracy using features received in the cloud. To achieve higher feature compression performance, redundancy between feature channels can be eliminated through feature rearrangement, feature transformation, and feature quantization. Summary of the Invention

[0004] In view of the problems existing in the current visual laser data feature compression and extraction system, this invention is proposed.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] In a first aspect, embodiments of the present invention provide a method for compressing and extracting visual laser data features, which includes the following steps:

[0007] Establish a feature compression framework based on convolutional neural networks to extract intermediate features;

[0008] Based on the statistical characteristics of the features, identify redundant or low-information non-information features, and delete the features most similar to the non-information features.

[0009] The discrete cosine transform is introduced to further compress statistical redundancy, and redundancy between feature channels is eliminated through feature rearrangement, feature transformation and feature quantization.

[0010] In the receiver, the removed features are recovered based on the features of the non-informative feature subset.

[0011] As a preferred embodiment of the visual laser data feature compression and extraction method of the present invention, the specific steps further include:

[0012] Let the input image be x. The features obtained from the intermediate layers of network N form a feature tensor space with a size of H×C×W, where C, H, and W represent the number of channels, height, and width of the features, respectively. Before the nonlinear activation function, the output characteristic f(l) of the l-th layer is given by the formula:

[0013] f(l)=N(x,l)=∑h(l,i)*f(l-1,i)+α(l,i);

[0014] In the formula, h(l,i) represents the convolution parameter of the i-th channel in the l-th layer, α(l,i) is the bias parameter of the convolutional neural network, and * represents the 2D convolution operation;

[0015] Identify the features least relevant to the monitoring task and form a non-informative subset S. n Based on the similarity metric, the feature most similar to the non-informative feature is selected to form a similarity subset S. s S s and S n The features in the dataset have a one-to-one correspondence, and then all features are divided into three subsets;

[0016] S = S n ∪S i ∪S s ={f(l,i),i∈[1,C]};

[0017] In the formula, S represents the entire feature, S i It is the set of the remaining features. S s The features in the code will be removed from the encoder;

[0018] Output feature S ψ Represented as:

[0019] S ψ =Ψ(S);

[0020] In the receiver, it is possible to determine the value based on S. n The corresponding features in the model are used to restore the removed features;

[0021] S ψ The transformation is performed using Γ[·]. Finally, a discrete code y(l) is generated through quantization and entropy coding, denoted by C[·].

[0022] y(l)=C(Γ(S ψ ),θ C );

[0023] In the formula, θC This represents the set of encoder parameters to be tuned;

[0024] In the decoder, features are reconstructed:

[0025]

[0026] In the formula, It is the inverse process of the corresponding process in the encoder.

[0027] As a preferred embodiment of the visual laser data feature compression and extraction method of the present invention, the steps of extracting intermediate features and determining phase velocity include:

[0028] The intermediate features include: non-information features, target information features, and a subset of similarity features;

[0029] The subset of non-informative features is represented as: S n ={f(l,i dx )|i dx ∈1,K};

[0030] Among them, i dx It is an index of non-information feature channels;

[0031] Therefore, the similarity of non-information features is represented as:

[0032]

[0033] Where, ρ i,j ρ represents the similarity score between f(l,i) and f(l,j). i,j Determined by Pearson correlation coefficient, mutual information, or k2 norm;

[0034] ρ i,j =||f(i,j)-f(l,j)||2.

[0035] As a preferred embodiment of the visual laser data feature compression and extraction method of the present invention, the discrete cosine transform is expressed as follows:

[0036] z f (l)=A T f′(l)A;

[0037] In the formula, z f (l) represents the transformed feature data, A is the transformation matrix, [·] T It is the matrix transpose.

[0038] As a preferred embodiment of the visual laser data feature compression and extraction method of the present invention, the step of eliminating redundancy between feature channels includes,

[0039] Without loss of generality, uniform quantization is adopted, and n-bit quantization is performed as follows:

[0040]

[0041] In the formula, max(z) f (l)) and min(z) f (l)) represents the maximum and minimum values ​​in the feature data, [z] q (l)] H,W It is the quantized feature data. This indicates rounding down to the nearest integer. q (l) will be fed into the arithmetic coding engine to further eliminate its statistical redundancy.

[0042] In a preferred embodiment of the visual laser data feature compression and extraction method of the present invention, the step of recovering representative features includes:

[0043] The feature extraction decoder consists of: entropy decoding, inverse quantization, inverse transform, and feature recovery;

[0044] The inverse quantization process is represented as follows:

[0045]

[0046] in, It is dequantized feature data.

[0047] In a preferred embodiment of the visual laser data feature compression and extraction method of the present invention, the step of recovering representative features further includes:

[0048] The tiled feature image is obtained through the DCT coefficient matrix, specifically represented as follows:

[0049]

[0050] By decomposing the feature data into feature channels for recovery, the features removed from the encoder will be obtained by S. s The matching feature reconstruction in the process is specifically represented as follows:

[0051]

[0052] in, To restore representative features, Features removed during the encoding process are then reconstructed, and the restored features are transmitted to the remaining network to continue the data analysis task.

[0053] Secondly, embodiments of the present invention provide a visual laser data feature compression and extraction system, which includes a feature compression framework, a feature statistical analysis module, a feature quantization module, and a matching module;

[0054] The feature compression framework is based on a convolutional neural network and is used to extract intermediate features.

[0055] The feature statistical analysis module is used to identify redundant or low-information non-informative features and delete the features most similar to the non-informative features.

[0056] The feature quantization module is used to further compress statistical redundancy and eliminate redundancy between feature channels;

[0057] The matching module is used to determine the features least relevant to the monitoring task, forming a non-informative subset, and to select the features most similar to the non-informative features, forming a similarity subset.

[0058] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any step of the above-described visual laser data feature compression and extraction method.

[0059] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the above-described visual laser data feature compression and extraction method.

[0060] The beneficial effects of this invention are as follows: it effectively compresses intermediate features. It studies feature matching based on similarity, identifies the least important features in the monitoring task, selects the features most relevant to non-informative features, and then divides the total features into non-informative subsets, informative subsets, and similarity subsets. Features in the similarity subset are deleted from the encoder and recovered based on the non-informative subset in the decoder, significantly reducing the pressure on data transmission and storage while ensuring the integrity and effectiveness of feature information. Furthermore, this invention introduces a two-dimensional discrete cosine transform (DCT) to further compress feature redundancy, allowing feature information to be flattened and forwarded to the DCT module to concentrate feature values, significantly reducing the code rate. The application of this technology not only enhances the real-time monitoring capability of the digital twin model for equipment status but also provides more accurate and refined data support and basis for operation and maintenance decisions, thus laying a solid technical guarantee for the efficient, stable, and safe operation of power transmission and transformation equipment. Attached Figure Description

[0061] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0062] Figure 1 This is a flowchart of a method for compressing and extracting visual laser data features.

[0063] Figure 2 This is a structural diagram of a method for compressing and extracting visual laser data features. Detailed Implementation

[0064] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0065] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0066] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0067] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0068] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0069] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0070] Example 1

[0071] Reference Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides a method for compressing and extracting visual laser data features, including the following steps:

[0072] S1. Establish a feature compression framework based on convolutional neural networks to extract intermediate features.

[0073] The steps of extracting intermediate features and determining phase velocities include:

[0074] The intermediate features include: non-information features, target information features, and a subset of similarity features;

[0075] The selection of non-information features can be determined based on mutual information, correlation, statistics, etc. Since features with small variance generally contain fewer details and have less impact on the task, feature variance is used as the standard. The computational complexity of variance is also acceptable. Features are sorted in ascending order based on variance. The feature with the highest ranking is selected as the non-information feature. The value of is related to the network layer of the feature and the monitoring task;

[0076] The subset of non-informative features is represented as: S n ={f(l,i dx )|i dx ∈1,K};

[0077] Among them, i dx It is an index of non-information feature channels;

[0078] Therefore, the similarity of non-information features is represented as:

[0079]

[0080] Where, ρ i,j ρ represents the similarity score between f(l,i) and f(l,j). i,j Determined by Pearson correlation coefficient, mutual information, or l2 norm;

[0081] ρ i,j =||f(i,j)-f(l,j)||2.

[0082] The number of features is large, and computing the entire similarity matrix is ​​very time-consuming. To simplify similarity matching, only the similarity between non-informative features and other features is calculated. That is, f(l,j)∈S n ,f(l,j)∈SS n , i≠j, similarity matrix It can be simplified to K×(C–K) for any feature f(l,i) dx )∈S n Matching similar features can be represented as an optimization problem:

[0083]

[0084] Where j * Is it related to f(l,i) dx The index of the matching feature with the highest similarity. In this way, S can be determined. n Find the similar features of each feature and form a similarity subset S. s .

[0085] S s ={f(l,j * )|j * ∈[1,C]}

[0086] Therefore S n and S s It is a one-to-one relationship and does not belong to S. n and S s The remaining features are assigned to subset S i To reduce the amount of data, S will be removed from the output features. s In the decoder, these deleted features will be processed through S n The matched features from the first layer are copied to recover the feature image. Finally, the features from the first layer are tiled to form the feature image x. f (l)={f′(l,i|f′(l,i)∈S i ∪S s )}.

[0087] S2. Based on the statistical characteristics of the features, identify redundant or low-information non-information features, and delete the features most similar to the non-information features.

[0088] The discrete cosine transform is expressed as follows:

[0089] z f (l)=A T f′(l)A;

[0090] In the formula, z f (l) represents the transformed feature data, A is the transformation matrix, [·] T It is the matrix transpose.

[0091] Discrete cosine transform introduces additional complexity. Typically, complexity increases with block size. Smaller block sizes generally have lower complexity but are more prone to boundary jumps. This patent transforms the entire feature data, resulting in more concentrated data, which is beneficial for subsequent entropy coding; therefore, the overall complexity does not increase significantly.

[0092] S3. The discrete cosine transform is introduced to further compress statistical redundancy. Redundancy between feature channels is eliminated through feature rearrangement, feature transformation and feature quantization.

[0093] The steps to eliminate redundancy between feature channels include,

[0094] Without loss of generality, uniform quantization is adopted, and n-bit quantization is performed as follows:

[0095]

[0096] In the formula, max(z) f (l)) and min(z) f (l)) represents the maximum and minimum values ​​in the feature data, [z] q (l)] H,W It is the quantized feature data. This indicates rounding down to the nearest integer. q (l) will be fed into the arithmetic coding engine to further eliminate its statistical redundancy.

[0097] S4. In the receiver, recover the removed features based on the features of the non-informative feature subset.

[0098] The steps for restoring representative features include:

[0099] The feature extraction decoder consists of: entropy decoding, inverse quantization, inverse transform, and feature recovery;

[0100] The inverse quantization process is represented as follows:

[0101]

[0102] in, It is dequantized feature data.

[0103] The steps for restoring representative features also include,

[0104] The tiled feature image is obtained through the DCT coefficient matrix, specifically represented as follows:

[0105]

[0106] By decomposing the feature data into feature channels for recovery, the features removed from the encoder will be obtained by S. s The matching feature reconstruction in the process is specifically represented as follows:

[0107]

[0108] in, To restore representative features, Features removed during the encoding process are then reconstructed, and the restored features are transmitted to the remaining network to continue the data analysis task.

[0109] The specific steps also include,

[0110] In non-generative feedforward networks, consecutive layers form a Markov chain. With fixed network weights, the output feature of any layer is uniquely determined by the input through feedforward processing. Convolutional layers have multiple convolutional kernels, which generate corresponding feature channels.

[0111] Let the input image be x. The features obtained from the intermediate layers of network N form a feature tensor space with a size of H×C×W, where C, H, and W represent the number of channels, height, and width of the features, respectively. Before the nonlinear activation function, the output characteristic f(l) of the l-th layer is given by the formula:

[0112] f(l)=N(x,l)=∑h(l,i)*f(l-1,i)+α(l,i);

[0113] In the formula, h(l,i) represents the convolution parameter of the i-th channel in the l-th layer, α(l,i) is the bias parameter of the convolutional neural network, and * represents the 2D convolution operation;

[0114] Identify the features least relevant to the monitoring task and form a non-informative subset S. n Based on the similarity metric, the feature most similar to the non-informative feature is selected to form a similarity subset S. s S s and S nThe features in the dataset have a one-to-one correspondence, and then all features are divided into three subsets;

[0115] S = S n ∪S i ∪S s ={f(l,i),i∈[1,C]};

[0116] In the formula, S represents the entire feature, S i It is the set of the remaining features. S s The features in the code will be removed from the encoder;

[0117] Output feature S ψ Represented as:

[0118] S ψ =Ψ(S);

[0119] In the receiver, it is possible to determine the value based on S. n The corresponding features in the model are used to restore the removed features;

[0120] S ψ The transformation is performed using Γ[·]. Finally, a discrete code y(l) is generated through quantization and entropy coding, denoted by C[·].

[0121] y(l)=C(Γ(S ψ ),θ C );

[0122] In the formula, θ C This represents the set of encoder parameters to be tuned;

[0123] In the decoder, features are reconstructed:

[0124]

[0125] In the formula, It is the inverse process of the corresponding process in the encoder.

[0126] In summary, this method effectively compresses intermediate features. A similarity-based feature matching approach was studied to identify the least important features in the monitoring task, select the features most relevant to the non-informative features, and then divide the total features into non-informative, informative, and similarity subsets. Features in the similarity subset are deleted from the encoder and recovered based on the non-informative subset in the decoder, significantly reducing the pressure on data transmission and storage while ensuring the integrity and effectiveness of feature information. Furthermore, this invention introduces a two-dimensional discrete cosine transform (DCT) to further compress feature redundancy, allowing feature information to be flattened and forwarded to the DCT module to concentrate feature values, significantly reducing the bit rate. The application of this technology not only enhances the real-time monitoring capability of the digital twin model for equipment status but also provides more accurate and refined data support and basis for operation and maintenance decisions, thus laying a solid technical guarantee for the efficient, stable, and safe operation of power transmission and transformation equipment.

[0127] Example 2

[0128] Based on the first embodiment, this embodiment further provides a visual laser data feature compression and extraction system, including a feature compression framework, a feature statistical analysis module, a feature quantization module, and a matching module;

[0129] The feature compression framework is based on a convolutional neural network and is used to extract intermediate features.

[0130] The feature statistical analysis module is used to identify redundant or low-information non-informative features and delete the features most similar to the non-informative features.

[0131] The feature quantization module is used to further compress statistical redundancy and eliminate redundancy between feature channels;

[0132] The matching module is used to determine the features least relevant to the monitoring task, forming a non-informative subset, and to select the features most similar to the non-informative features, forming a similarity subset.

[0133] This embodiment also provides a computer device applicable to the visual laser data feature compression and extraction method, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the visual laser data feature compression and extraction method proposed in the above embodiment.

[0134] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0135] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the method for compressing and extracting visual laser data features as proposed in the above embodiments.

[0136] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for compressing and extracting visual laser data features, characterized in that: Includes the following steps, Establish a feature compression framework based on convolutional neural networks to extract intermediate features; Based on the statistical characteristics of the features, identify redundant or low-information non-information features, and delete the features most similar to the non-information features. The discrete cosine transform is introduced to further compress statistical redundancy, and redundancy between feature channels is eliminated through feature rearrangement, feature transformation and feature quantization. In the receiver, the removed features are recovered based on the features of the non-informative feature subset; The steps to eliminate redundancy between feature channels include, Without loss of generality, uniform quantization is adopted, and n-bit quantization is performed as follows: In the formula, max(z) f (l)) and min(z) f (l)) represents the maximum and minimum values ​​in the feature data, [z] q (l)] H,W It is the quantized feature data. This indicates rounding down to the nearest integer, z q (l) will be fed into the arithmetic coding engine to further eliminate its statistical redundancy; The steps for restoring representative features include: The feature extraction decoder consists of: entropy decoding, inverse quantization, inverse transform, and feature recovery; The inverse quantization process is represented as follows: in, It is dequantized feature data; The steps for restoring representative features also include, The tiled feature image is obtained through the DCT coefficient matrix, specifically represented as follows: By decomposing the feature data into feature channels for recovery, the features removed from the encoder will be obtained by S. s The matching feature reconstruction in the process is specifically represented as follows: in, To restore representative features, Features removed during the encoding process are then reconstructed, and the restored features are transmitted to the remaining network to continue the data analysis task.

2. The method for compressing and extracting visual laser data features as described in claim 1, characterized in that: The specific steps also include, Let the input image be x. The features obtained from the intermediate layers of network N form a feature tensor space with a size of H×C×W, where C, H, and W represent the number of channels, height, and width of the features, respectively. Before the nonlinear activation function, the output characteristic f(l) of the l-th layer is given by the formula: f(l)=N(x,l)=∑h(l,i)*f(l-1,i)+α(l,i); In the formula, h(l,i) represents the convolution parameter of the i-th channel in the l-th layer, α(l,i) is the bias parameter of the convolutional neural network, and * represents the 2D convolution operation; Identify the features least relevant to the monitoring task and form a non-informative subset S. n Based on the similarity metric, the feature most similar to the non-informative feature is selected to form a similarity subset S. s S s and S n The features in the dataset have a one-to-one correspondence, and then all features are divided into three subsets; S=S n ∪S i ∪S s ={f(l,i),i∈[1,C]}; In the formula, S represents the entire feature, S i It is the set of the remaining features, S s The features in the code will be removed from the encoder; Output feature S ψ Represented as: S ψ =Ψ(S); In the receiver, it is possible to determine the value based on S. n The corresponding features in the model are used to restore the removed features; S ψ The transformation is performed by Γ[·], and finally, a discrete code y(l) is generated by quantization and entropy encoding, denoted by C[·]. y(l)=C(Γ(S ψ ),θ C ); In the formula, θ C This represents the set of encoder parameters to be tuned; In the decoder, features are reconstructed: In the formula, It is the inverse process of the corresponding process in the encoder.

3. The method for compressing and extracting visual laser data features as described in claim 2, characterized in that: The steps of extracting intermediate features and determining phase velocities include: The intermediate features include: non-information features, target information features, and a subset of similarity features; The subset of non-informative features is represented as: S n ={f(l,i dx )|i dx ∈1,K}; Among them, i dx It is an index of non-information feature channels; Therefore, the similarity of non-information features is represented as: Where, ρ i,j ρ represents the similarity score between f(l,i) and f(l,j). i,j Determined by Pearson correlation coefficient, mutual information, or l2 norm; r i,j =||f(i,j)-f(l,j)||2.

4. The method for compressing and extracting visual laser data features as described in claim 3, characterized in that: The discrete cosine transform is expressed as follows: z f (l)=A T f′(l)A; In the formula, z f (l) represents the transformed feature data, A is the transformation matrix, [·] T It is the matrix transpose.

5. A visual laser data feature compression and extraction system, based on the visual laser data feature compression and extraction method according to any one of claims 1 to 4, characterized in that: It includes a feature compression framework, a feature statistical analysis module, a feature quantization module, and a matching module; The feature compression framework is based on a convolutional neural network and is used to extract intermediate features. The feature statistical analysis module is used to identify redundant or low-information non-informative features and delete the features most similar to the non-informative features. The feature quantization module is used to further compress statistical redundancy and eliminate redundancy between feature channels; The matching module is used to determine the features least relevant to the monitoring task, forming a non-informative subset, and to select the features most similar to the non-informative features, forming a similarity subset.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the visual laser data feature compression and extraction method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the visual laser data feature compression and extraction method according to any one of claims 1 to 4.

Images